Changes proposed in this Pull Request

Land transport is split from a single passenger-car aggregate into five motor
vehicle segments:

1. passenger cars (pc),
2. powered two-wheelers (ptw),
3. buses/coaches (bus),
4. light commercial vehicles (lcv), and
5. heavy goods vehicles (hgv)

so that fleet size, driving profile, propulsion-mix shares,
charging behaviour and EV characteristics can differ per segment.

• Demand build (build_transport_demand.py) is now km-based per segment. Total
  vehicle-km per country come from JRC-IDEES road (and, for bus/hgv, a share
  of non-electrified rail moved onto road via load-factor ratios); Propulsion-based
  correction of energy totals can be removed hence; the weekly demand shape derives
  from per-segment BASt profiles.
• BASt profiles (build_mobility_profiles.py) now also produce mot, bus,
  lfw and lkw weekly profiles alongside kfz/pkw.
• Energy totals (build_energy_totals.py) extract per-segment vehicle stock,
  vehicle-km driven, load factors and an hgv efficiency from IDEES road and rail
  activity sheets.
• Network build (prepare_sector_network.py) loops over segments and builds a
  segment-suffixed EV battery bus, load, BEV charger, DSM store and V2G link for
  each. transport_electric_efficiency for bus/hgv is taken from
  technology-data. low_voltage_charging selects which segments connect to the
  LV distribution grid.
• Config: BEV/EV scalars and the propulsion-mix share dicts become
  per-segment mappings; new options bev_dem_smoothing (was the previously
  unconditional EV power smoothing) and low_voltage_charging. Pydantic models
  in scripts/lib/validation/config/sector.py and config/schema.default.json
  updated to accept value | {segment: value}.
• Plotting: tech_colors for the per-segment EV battery, land transport demand, land transport EV, BEV charger and V2G carriers.

Status: this is a work in progress

Working end-to-end for the electric segments. Known gaps before this is
merge-ready are listed in the collapsed section below; the headline ones:

1. No aggregation switch yet. The pipeline is hard-wired to five segments.
   The intended pre-commit state is a config switch that (a) collapses all
   segments into the kfz aggregate as today's default behaviour, (b) restricts
   DSM to pkw only in that mode, while (c) keeping the new km-based demand
   inputs. That switch is not implemented.
2. ICE and fuel-cell segments are not disaggregated. add_ice_cars and
   add_fuel_cell_cars are still called with the pc share only (marked
   # TEMP: pc) and currently sit inside the per-segment loop, so they misbehave
   for a multi-segment run. Needs either a clean aggregate call outside the loop
   or a proper per-segment treatment.

Checklist

Required:

• Changes are tested locally and behave as expected.
• Code and workflow changes are documented.
• A release note entry is added to doc/release_notes.rst.
• The description is human-written and any AI-generated content is marked.

If applicable:

• Changes in configuration options are reflected in scripts/lib/validation.
• For new data sources or versions, these instructions have been followed.
• New rules are documented in the appropriate doc/*.rst files.

Notes on the checklist: the config validation models are updated. No new rules were added (only new outputs on build_mobility_profiles and new inputs on build_transport_demand). The BASt source is unchanged: only additional columns are read. Release note and doc/ updates are pending.

Implementation notes — data flow, per-segment wiring, and open gaps (verbose, AI-assisted draft)

1. What changed, stage by stage

1a. Weekly profiles — build_mobility_profiles.py

BASt hourly counting-point data is aggregated to weekly profiles for the five
segments in addition to the existing kfz (all motor vehicles) and pkw:

The rule (rules/build_sector.smk) gains mot/bus/lfw/lkw outputs and
build_transport_demand gains the matching traffic_data_Mot/Bus/Lfw/Lkw
inputs. MOBILITY_PROFILES_DATASET["folder"] is now wrapped in Path(...) so
the / operator works.

1b. Energy / activity totals — build_energy_totals.py

New columns pulled from JRC-IDEES TrRoad_act, TrRoad_ene, TrRail_act,
TrRail_ene:

• Number <segment>: vehicle stock for all five segments (was: passenger cars
  only).
• mio km-driven <segment>: road vehicle-km for all five segments.
• mio km-driven Rail / Rail passenger / Rail freight.
• load factor for Rail passenger, Rail freight, Heavy goods vehicles,
  Motor coaches, buses and trolley buses.
• heavy goods efficiency (kgoe/100km → ktoe/100km, like the existing
  passenger car efficiency).

Unit handling: Number*, mio km-driven* and load factor* are excluded from
the ktoe→TWh scaling (str.fullmatch pattern list); efficiency columns are
divided by 1e6. Gap-filling now loops over car_cols / stats instead of the
single number cars / average fuel efficiency series. The renamed output
column is Number Passenger cars (old number cars is gone).

Gap / risk

• The Swiss patch only renames passenger cars → Number Passenger cars; the
  other four Number * / mio km-driven * columns are NaN for CH and rely
  entirely on the per-column averaged gap-fill. Confirm that is acceptable.

1c. Demand time series — build_transport_demand.py

build_transport_demand was rewritten:

• get_shape(traffic_fn) factored out: returns the normalised weekly shape.
• Per-segment total demand in mio km:
  • pc, ptw, lcv: straight from mio km-driven <segment>.
  • bus: mio km-driven buses plus non_elec_rail · mio km-driven Rail passenger · (load factor Rail passenger / load factor bus).
  • hgv: mio km-driven HGV plus non_elec_rail · mio km-driven Rail freight · (load factor Rail freight / load factor HGV), where
    non_elec_rail = 1 − electricity rail / total rail.
• get_demand(profile, total, nyears, seg) → profile · total · 1e4 · nyears,
  wrapped in a MultiIndex column level keyed by segment.
• Output is a wide frame with a two-level column index (segment, node).
• bev_availability_profile and bev_dsm_profile likewise produce
  (segment, node) columns; availability reads bev_avail_max[seg] /
  bev_avail_mean[seg], DSM reads bev_dsm_restriction_time[seg] /
  bev_dsm_restriction_value[seg].

Behaviour change worth calling out

• The ICE heating/cooling demand correction (dd_ICE, ice_correction) and
  the whole temperature/transport_degree_factor block were removed from the
  demand build. Demand is now purely km-based; temperature correction survives
  only inside add_EVs via get_temp_efficency. FC and ICE demand no longer get
  the build-stage correction at all.
• airtemp_fn is still a parameter of build_transport_demand but is unused.

1d. Network — prepare_sector_network.py

• add_EVs gains car_efficiency and transport_type parameters. Every added
  component is suffixed with the segment: EV battery {t}, land transport EV {t}, BEV charger {t}, EV battery {t} store, V2G {t}. Carriers are
  segment-specific too.
• car_efficiency is passed in (no longer read from options inside add_EVs).
• The rolling-average power smoothing in add_EVs is now gated on
  options["bev_dem_smoothing"] (previously unconditional).
• get_car_efficiencies() builds a DataFrame of 100km/MWh efficiencies:
  pc/ptw/lcv from options["transport_electric_efficiency"][seg]; bus/hgv
  from technology-data (costs.loc[car_keys[engine][seg], "efficiency"].mean(),
  converted from kWh/km). engine_types is hard-coded to ["electric"] only.
• add_land_transport: reads the three files with header=[0, 1], reindexes to
  nodes on column level 1, loops for transport_type in transport_types and
  calls add_EVs(...) per segment.
• insert_electricity_distribution_grid: BEV chargers / V2G are moved to
  " low voltage" only for carriers matching
  BEV charger (pc|ptw|bus|lcv|hgv) filtered by options["low_voltage_charging"].
• check_land_transport_shares now takes the shares DataFrame and warns per
  column whose propulsion shares do not sum to 1.

1e. Config + validation

• scripts/lib/validation/config/sector.py: _TRANSPORT_SEGMENTS and
  _segment_share_default() helpers; the affected fields become
  float | dict[str, float] / bool | dict[str, bool] /
  dict[int, float] | dict[str, dict[int, float]] with per-segment
  default_factory values.
• bev_dsm (old scalar bool, "participate in DSM") keeps its name and meaning
  but is now per-segment. bev_dem_smoothing is a new key that took over the
  power-profile smoothing which used to be unconditional.
• Numeric defaults changed for existing single-value options, e.g.
  bev_dsm_availability 0.5 → ~0.33–0.45, bev_charge_rate 0.011 → 0.0266 (pc),
  transport_electric_efficiency 53.19 → 80.0 (pc). Needs a release note with the sources for the new numbers
• config/schema.default.json mirrors the pydantic changes with anyOf
  branches; both copies (sector and the scenario-override block) are updated.

2. Data-flow summary

BASt hourly counts ──build_mobility_profiles──► kfz, pkw, mot, bus, lfw, lkw (weekly shape)
JRC-IDEES road+rail ─build_energy_totals──────► Number <seg>, mio km-driven <seg>,
 mio km-driven Rail{,passenger,freight},
 load factor {Rail p/f, HGV, bus},
 passenger car / heavy goods efficiency
 │
 ├─ build_transport_data ──► transport_data.csv (per-country stock, load factors,
 │ average fuel efficiency)
 ▼
build_transport_demand
 nodal split by population ─► nodal_transport_data
 per-segment total mio km ─► + rail-passenger→bus, rail-freight→hgv (non-elec share only)
 ×ばつ weekly shape ×ばつ 1e4 ×ばつ nyears
 ├─► transport_demand_s_{clusters}.csv columns = (segment, node), unit [100 km]
 ├─► avail_profile_s_{clusters}.csv columns = (segment, node)
 └─► dsm_profile_s_{clusters}.csv columns = (segment, node)
 ▼
prepare_sector_network.add_land_transport
 for segment in {pc, ptw, bus, lcv, hgv}:
 shares[engine, segment] from land_transport_<engine>_share[segment]
 electric > 0 ─► add_EVs(..., car_efficiency=get_car_efficiencies()[electric, segment],
 transport_type=segment)
 EV battery {seg} bus + load + BEV charger {seg}
 (+ EV battery {seg} store + V2G {seg} if bev_dsm[seg] / v2g[seg])
 fuel_cell > 0 ─► add_fuel_cell_cars(... pc share ...) ◄── TEMP, not segmented
 ice > 0 ─► add_ice_cars(... pc share ...) ◄── TEMP, not segmented
 insert_electricity_distribution_grid:
 BEV charger {seg} / V2G {seg} → " low voltage" if seg in low_voltage_charging

3. The aggregation switch (not yet implemented)

Target design for the pre-commit state:

• New config flag, e.g. sector.land_transport_segmentation: false (default),
  choosing between:
  • aggregated: one segment, the kfz weekly shape, one set of scalar
    parameters, demand still taken from the km-based totals summed over segments
    (overwriting the legacy energy-based path).
  • segmented: the five-segment path in this PR.
• In aggregated mode, DSM is added for pkw only (the historical assumption),
  i.e. the DSM store / V2G are built for the passenger-car equivalent and skipped
  for the rest.
• Both modes consume the same km-based inputs so the demand build does not fork.

Until that exists, config.default.yaml forces the five-segment path on every
run.

4. ICE / FC segmentation (open question)

car_keys in prepare_sector_network.py already lists technology-data cost
entries for fuel_cell and ice bus/hgv vehicles, so the efficiency lookup
generalises. What is missing:

• get_car_efficiencies() limited to engine_types = ["electric"]; extend to
  fuel_cell / ice and feed the per-segment efficiency into
  add_fuel_cell_cars / add_ice_cars (both need a car_efficiency /
  transport_type parameter, mirroring add_EVs).
• Decide the modelling scope: do ptw / bus / lcv / hgv get their own
  H2 / oil demand buses and carriers, or is FC/ICE kept aggregate while only EV
  is segmented? The former multiplies carriers by 5; the latter needs the FC/ICE
  calls pulled out of the per-segment loop and fed a summed p_set and a
  fleet-weighted share.

5. Pre-review cleanup checklist

• Aggregation switch (kfz fallback, pkw-only DSM) implemented and default.
• Scalar config values normalised to per-segment dicts at load time (or the
  union / schema text dropped).
• FC / ICE calls moved out of the per-segment loop; decision on FC/ICE
  segmentation recorded.
• get_car_efficiencies() takes options / costs as parameters.
• Re-wire traffic_data_KFZ rule input parameterfor the aggregation switch.
• Sources / citations for the new numeric defaults in the release note.
• doc/ updates: configuration.md (new + changed options), sector.md
  transport section, preparation.md / rules doc for the new profile outputs.
• Release note entry in doc/release_notes.md (breaking config changes:
  bev_dsm semantics, bev_dem_smoothing rename, per-segment defaults).